Plea to Server Operators

[vonunov]

We all appreciate people who use their time and resources to provide AC servers, but you must make sure you don't defeat the purpose of that effort. I see many 20-or-so slot servers that simply cannot support their capacity.

Please read The Complete Idiot's Guide to AC Servers (PDF). Though it was written for an old version, it still contains valuable information, such as bandwidth requirements.



Note that the UPLOAD speed requirement for a 20-slot server is about 3Mb. Many home connections do not get 3Mb upstream. On a home server, you must also take into account the other usage of the network. If you get 3Mb upstream, that doesn't mean you can run a 20-slot server, unless you don't use that internet connection for anything else.

Note, by the way, that a 20-slot server transfers over a terabyte of data per month. Something to consider if your connection is metered.

Lastly, check in on your servers every once in a while, especially when running at the configured full capacity, to check that everything runs well. If you're getting serverlag, it's time to reconsider your capacity.

[gambino]

Thanks this is very helpful, it seems I am able to host 20 man servers, but imo 10v10 on any map is too much. I think servers should be 12 slots max......

[Ronald_Reagan]

I would suggest 16 slot max :P

[#M|A#Wolf]

I agree with RR. I can't take the 20 player games now (don't really know how I was able to play the 50 player bs_dust2 servers, even though they were AWESOME), and 16 would be a perfect limit.

[Ronald_Reagan]

(except for -mlocalhost servers)

http://www.speedtest.net/ this will help you figure out your upload and download.

[flowtron]

Well - the 20-max limit resulted in a giant wave of indignation by some inside the community when it was enforced - seems the posters in this thread understand what drove the dev-team to impose this limit at the time. The way the game is meant to be played.
I totally agree re:10-vs-10 is too much for most maps, or a 20 player DM, however you like it. For the record, I feel that the game works best with teamsizes ranging from 3 to 6 - but this utopian-view only really holds true if we're talking about teams that actually play as such and aren't just randomly assigned .. as is the case on 87.3% of public servers ;-)

Re: the bandwidth table - AFAIK nobody has yet updated this to the protocol-slimming efforts Brahma has put into the last release! The figures are a good orientation still, but they're not accurate anymore!

I'd plea for 3 servers for each 20-slot you have running: 1x 1-vs-1/coop, 1x 3-vs-3 and 1x 6-vs-6 would be a nice way to go (or 2x 3-vs-3 and 1x 4-vs-4, or or or) .. of course you're still free to host 20-slot servers with maprotations full of custom maps.
But wouldn't more specific setups maybe help alleviate the heavy fluctuation I always see - when I find the time to actually play the game :-P - people joining and leaving all the time. You sometimes manage to have a couple of matches in a row with a few people, naturally more often the better you know them .. and the better they like you! ..
If you provide a low-slot-count server with a CTF maprotation, a server-title with CTF in it (up front, not after 20+ chars of "I'm the man!" *ggg*) you're bound to get people actually wanting to play CTF!

[Ronald_Reagan]

There is a need for some servers to host more spots for spectators. This is why I mentioned the -m switch.

[DES|Anderson]

I personally believe 12 or 14 works well, also dont bash home server ops in my experience a sensibly run home server is no worse than one run on VPS. With regard to lag there is more at play here, I experience lag issues on my servers and i thought maybe it was capacity but recently I have been logging trffic levels ect and trying to tie it up with lag and it does not corelate! StrangeI know, so i am trying to find out what else is going on. The lag I experience is fleeting like someone took all the resources away for a second or two, its no worse on mine than any other server out there infact I believe mine to be better than most. What about lag on the MS, has anyone looked into that or have we just assumed its always the server itself, dont all troll me out I dont know fully the role of the MS durring game (if any) it was just a sugestion.

[vonunov]

I'm pretty sure the MS just indexes the servers that are online. AC traffic isn't all routed through the MS.

Not bashing home servers ops, either, just the ones who think they can run a 20-slot server on their 1.5Mb upstream DSL. ;)

[pwnage{TyD}]

Those numbers in the first post were reduced 30% in the SVN before 1.1 was released. They're not correct up to date, but gives a good indication of what line you must have in order to host 2x14 slots servers, or more.

[jamz]

You're right, Jack, the MS is just an index of servers.
The coloured bars on the in-game server list give a good indication of what your experience on a server will be like. Servers that are all red or red/yellow are going to give you problems at a certain number of players. Yellow or yellow/green is normal. All green is an extra-special good server.

A poor connection might not just be the bandwidth being exceeded, but could be the actual line quality, ie. too many lost packets. It is nothing to do with ping, therefore distance from the server is irrelevant.
You can find out what the approximate bandwidth capacity will be by pressing F9 followed by F2 for any server.

The -[pRo]- server has a lot of errors up to 8 players, but then gets much worse at 9 to 11, then gives constant errors from 12 players up. The owner of this server, if he doesn't just forget it altogether, should limit the server to 8 players.

[vonunov]

Will that only display properly during or after a period of actual traffic?

True, it's not always caused by exceeding bandwidth. There are lower-capacity servers with bad connections and all that. I'm just here to complain about the ones -- like -[pRo]- -- who shouldn't have their capacity nearly that high and would probably better serve the community by *not* having a[nother] 20-slot running.

Is there a list of current bandwidth requirements or will I have to fire up some bandwidth monitoring tools and do it myself? :P

[tempest]

Will that only display properly during or after a period of actual traffic?

AFAIK the data is collected during the match and discarded at the end, so it should become more accurate the longer the match has already taken, but have nothing to do with how long the server has been running.

Is there a list of current bandwidth requirements or will I have to fire up some bandwidth monitoring tools and do it myself? :P

Well, I guess that was how the data we have currently was determined, so yes ;) I doubt Brahma can give you any exact values.

[jamz]

Sorry tempest, it is cumulative. Server owners can see the data if they have their server logging at debug level, under 'Uplink quality'.
If you're going to log the data transfer Jack, remember that different modes transfer different amounts of data; you have only to look at demo sizes to see this. This means that a server running mostly T/OSOK is going to require a lot less bandwidth than a server with lots of bullets and grenades flying about.

[Ronald_Reagan]

This means that a server running mostly T/OSOK is going to require a lot less bandwidth than a server with lots of bullets and grenades flying about.

... thats how noobs can get away with this.

Anyways, I'd say keep using that graph. It is always good to have a bit of head room. And it might knock some servers down a few more notches :>

[vonunov]

I have debug on and noticed the Uplink Quality lines, but never got around to interpreting them. Good stuff, though, now that I get it.

Thanks for the tip on modes, I'll keep that in mind.

[DES|Anderson]

Hmmm, awesom info and very helpfull I can see on my servers i have a igh error rate but it stays constant in fact as a ratio of time played it gets better when more play???

How do I address this issue and what do you think it might be?

[tempest]

Sorry tempest, it is cumulative.

Whoops, indeed, got confused there. *

* blame it on the code! :P

[Joe Smith]

" like -[pRo]- -- who shouldn't have their capacity nearly that high and would probably better serve the community"

^ should be the focus, i also agree that the max slots should be 14-16 on public servers ( theres no need for more than that, plus maps are not big enough for that many people anyway ) and 20 on private servers ( for spec's)

[Viper]

Regarding the total monthly bandwidth, does that assume that you have (x) number clients in your server 24/7 the entire month? i.e. no time when there might not be any people in the server?

[vonunov]

I think it assumes that constant capacity; i.e., if you run a 20-slot, you should be prepared to consume over a terabyte of bandwidth per month, even if it won't actually be full all day, every day.

[Viper]

The reason I ask this is because while I have sufficient speed


(wireless :D)

My ISP meters my transfer at 250 gb -_-

[tempest]

I'm sure they supply some kind of statistics where you see how much you already used up, so you could just run the server for a few days, then check again.

PS: I guess their TOS actually prohibit you from hosting a server, right? ;)
PPS: I just read 'wireless' there. Please test the connection's stability and jitter. That's the problem with all sorts of wireless (and AC), not throughput or (constant) ping.

[Viper]

No I've hosted for a while but i almost reached my limit a few months ago so i stopped :P

edit: I'm not new to this [servers] I just wanted to know if I could realisticly interpret that graph.